Filing gate: ① defect. Class (b) — a published, customer-readable page documents refusals the runtime never produces.
Filed by the domain:devx execution seat (domain:devx#1, session_01VDtqoecgES7ScQYGbFVDRv) from an out_of_scope_findings entry of the #19848 round (PR #19878), re-measured by this seat on origin/main before filing. ⛔ Ungraded and unrouted — triage's.
The carrier
content/docs/api/error-catalog.mdx, the ### VALUE_TOO_LONG entry (~:204) and the VALUE_TOO_SHORT entry beside it: each gives a live cause and a fix, as if a client should branch on the code.
What contradicts it
- No producer. Outside tests, the only hits for either code on
origin/main are the enum members themselves, packages/spec/src/api/errors.zod.ts:58-59.
- What the runtime answers instead. A length miss on a record is a field-level
max_length / min_length entry inside VALIDATION_FAILED (packages/objectql/src/validation/record-validator.ts:695-699); a Zod-parsed body maps too_big / too_small the same way (packages/spec/src/api/zod-issues-to-fields.ts:78-81).
- The declared contract: ADR-0114 keeps field-shaped top-level members only as a wart; the useful answer is the
fields[] entry.
⇒ A client that follows the page branches on a code it never receives.
Sibling, ⛔ not the same card
#19848 (PR #19878) repairs the INVALID_FORMAT entry on the same page the same way — "no route emits it; branch on VALIDATION_FAILED + fields[].code". That PR's file surface is the INVALID_FORMAT entry; these two entries are a separate edit. VALUE_OUT_OF_RANGE (1 producer) and MISSING_REQUIRED_FIELD (3 producers) do have emitters and are ⛔ not claimed here.
Dedupe
REST enumeration of open + closed issues updated since 2026-08-15 (⛔ /search is not reachable from this seat), title + body grepped for VALUE_TOO_LONG|VALUE_TOO_SHORT|error-status-unpinned: 0 hits. Radius: issues touched in that window; an older untouched card would be outside it.
Dedupe words: VALUE_TOO_LONG · VALUE_TOO_SHORT · error-catalog unemitted code · max_length field-level
Generated by Claude Code
Filing gate: ① defect. Class (b) — a published, customer-readable page documents refusals the runtime never produces.
Filed by the
domain:devxexecution seat (domain:devx#1,session_01VDtqoecgES7ScQYGbFVDRv) from anout_of_scope_findingsentry of the #19848 round (PR #19878), re-measured by this seat onorigin/mainbefore filing. ⛔ Ungraded and unrouted — triage's.The carrier
content/docs/api/error-catalog.mdx, the### VALUE_TOO_LONGentry (~:204) and theVALUE_TOO_SHORTentry beside it: each gives a live cause and a fix, as if a client should branch on the code.What contradicts it
origin/mainare the enum members themselves,packages/spec/src/api/errors.zod.ts:58-59.max_length/min_lengthentry insideVALIDATION_FAILED(packages/objectql/src/validation/record-validator.ts:695-699); a Zod-parsed body mapstoo_big/too_smallthe same way (packages/spec/src/api/zod-issues-to-fields.ts:78-81).fields[]entry.⇒ A client that follows the page branches on a code it never receives.
Sibling, ⛔ not the same card
#19848 (PR #19878) repairs the
INVALID_FORMATentry on the same page the same way — "no route emits it; branch onVALIDATION_FAILED+fields[].code". That PR's file surface is theINVALID_FORMATentry; these two entries are a separate edit.VALUE_OUT_OF_RANGE(1 producer) andMISSING_REQUIRED_FIELD(3 producers) do have emitters and are ⛔ not claimed here.Dedupe
REST enumeration of open + closed issues updated since 2026-08-15 (⛔
/searchis not reachable from this seat), title + body grepped forVALUE_TOO_LONG|VALUE_TOO_SHORT|error-status-unpinned: 0 hits. Radius: issues touched in that window; an older untouched card would be outside it.Dedupe words:
VALUE_TOO_LONG·VALUE_TOO_SHORT·error-catalog unemitted code·max_length field-levelGenerated by Claude Code